From Miles to “Electric Miles”: the Constantly Changing Musical Style of Jazz-trumpetist Miles Davis

From Miles to “Electric Miles”: the Constantly Changing Musical Style of Jazz-trumpetist Miles Davis

When Miles Davis was asked the question what he had done that was so important in his life at an award dinner at the White House in 1987 he answered “well, I’ve changed music five or six times”. Even though this may be exaggerated and said by a person with quite an attitude it fascinated me when I read this in an article in an issue of the Rolling Stone magazine from 1991, written just after the death of the great jazz trumpeter. While Davis is most famous for his earlier work (read for albums like Kind of Blue) his repertory is a lot bigger then this. Davis’ fascination for new instruments, technology and innovation comes back in the music he made, with which he influenced many other musicians. This is the reason why he is called “Electric Miles” at a certain point and why he lost a big part of his old fanbase (they found that Davis became too much involved in the world of pop music and therefore doubted his authenticity). But despite the negativity of his older fans it is exactly this interest I want to research. If it is true that Miles Davis included more and more electric elements (for example instruments, but also methods of recording) in his music overtime this should be visible in the statistics Spotify gives us (for example in the acousticness). I will set this off against the energy that is in the music because for me as listener this is variable changes a lot when I listen to tracks from different moments in Davis’ career.

For my research I’ve made a playlist on spotify with a selection of the most important albums Davis made during his life. This selection is not only based on the popularity of the albums, but also on where you can hear a clear change of style. I have not included every album he made because I’m convinced this will only make the results vaguer because of the overwhelming amount of data that will appear in the graphs. I’ve also excluded the work he made with composer Gil Evans, though this is an important part of his repertory. The reason I made this choice is because I see this cooperation more as a side project in which Miles Davis works with a big orchestra, something that he only does on the four albums he made with Evans in the period from 1957 until 1961.

Despite the exclusion of a certain part of his work I think the selection I made represents the oeuvre of Miles Davis well. In the end I hope to have a good and clear representation of the change of style in Davis’ music.

Miles Davis

Miles Davis

Findings

Acousticness against energy

In this graph you can see the acousticness of the music set of against the energy. The size of the dots is representing the popularity of a song. If you would draw a line you can see a decreasing tendency: if a song is less acoustic, its energy is higher and in reverse. It is only possible to say something about the results in relation to the flow of Davis’ career when you know in which order the albums appeared. That is why I’ve added a list of the albums in order of appearance here.

In the list of albums I’ve made a couple of clear divisions (divided by a space) These are the different periods in the work of Miles Davis, in which the music becomes less acoustic and more focused on popular features.

When you look at the songs from the albums in the first section you can see that they are mostly in the bottom section of the graph, with a high acousticness and a lower energy. The only exception in this section are the songs on the album Milestones, that are according to Spotify less acoustic and more energetic. That the songs are more energetic can be explained easily when you compare the songs to the other songs in this section. The tempo in most of them is much higher and when you listen to them they are in fact more energetic. Spotify’s analyses that the songs are less acoustic is more problematic, because there is no real reason to assume this is the case. In comparison with the other songs the same instruments are used and no computers or any other form of ‘electric’ instrument is used.

The second group of albums Davis made with his second great quartet. This quartet is famous because of drummer Wayne Shorter, who made a great deal of the compositions. From this you can conclude that the drums play a bigger role than before. For our results this could mean that the music is more energetic, but because of the lack of electrical influences on the music it should still be very acoustic. Although, when you look at the results this is not necessarily true. Only for Someday my Prince Will Come this is a correct observation; the rest of the tracks are scattered across the entire plot. Especially the tracks from Miles Smiles are according to the audio features Spotify gave us completely not acoustical and quite energetic. You could say that this is somewhat a grey area for Spotify, it is hard to analyze the rate of acousticness.

In the next group of albums Davis starts to experiment with electric features, like a band recorder and instruments like an electric (bass)guitar. Here you can see clearly that the centralization of the dots is shifting to the area of less acoustic and more energetic music. Only the tracks on Live - Evil are considered as more acoustic than the rest. An explanation for this is that the music is very experimental, and that the music - despite the use of the electric guitar and the Fender Rhodes - sounds acoustic because of the focus on percussion.

In the second to last album group you can see the influence of technological and musical innovations on the music of Davis very clearly. The albums in this category are described as Fusion, in which the use of popular instruments (which are by this time mostly electric) is very common. Most of the dots in the graph are in the energetic and non-acoustic region, what corresponds to my expectation of where they would be.

The tracks on the last album Davis made are completely based on a beat made on a drum machine and are clearly not acoustical, except for the use of voice and trumpet. One last time Davis changes his musical style radically into a fusion between jazz and hip hop, called acid jazz. Sadly there are no other albums he made in this genre to compare Doo-Bop with, because he died a few months after making it.

Timbre Analysis

Segmentation

On first sight timbre analysis doesn’t look very interesting for my research, but with this I want to point out the change in segmentation in Davis’ music. To show this I’ve visualized the self-similarity matrices of the tracks “Blue in Green”, “Nefertiti”, “Human Nature” and “The Doo-bop” below.

The resulting self-similarity matrices

The resulting self-similarity matrices

In the self-similarity matrix of “Blue in Green” you can see a very clear form of segmentation in the form of different solos. The bright parts are played by Davis himself on the trumpet. Its shrill sound is visualized as bright yellow, in contrast to the dark blue parts in which the piano plays the most important role. Around the middle you also see a vaguer (but still very clear) yellow part; John Coltrane’s saxophone solo. Most of Davis’ music (as usual in jazz) from this period is structured like this, in which of course the biggest part is played by himself.

The second self-similarity matrix (“Nefertiti”) has a much more complex structure, if it even is possible to speak of one. The composition is written by the drummer of Davis’ second great quintet, Wayne Shorter. On first hearing the track sounds very repetitive because the same melody is constantly repeated on trumpet and saxophone. However, this repetition isn’t clearly visible in the self-similarity matrix. An explanation for this is that Shorter is improvising on the drums in the back, which causes the lack of obvious segmentation. The yellow segments that are shown represent the moments where Shorter comes to a climax and where he really gets loose on the drums. So, although there is just one manner of structuration visible (climactic), you could speak of two (considering the repeated melody as the other).

“Human Nature” is a cover from Michael Jackons hit, so the structure is based on that of a popular song. This structure (verse-chorus) should be visible in the self-similarity matrix. For the analyzation it could also be interesting to compare the self-similarity matrix of Davis’ version of the song to the one of the original, shown below.

What makes it difficult to compare the self-similarity matrices is that the songs are built up differently. In a way they correspond, but on some points the segments differ in length. In both self-similarity matrices it is hard to distinguish verses from choruses, but what is clearly visible in both is the bridge and an unexpected alternation of the chorus near the end.

The timbre analysis of “The Doo-Bop Song” is mostly based on the speechiness of the different segments. A verse-chorus pattern is visible, but after a long intro from Davis on the trumpet.

Clustering the music

[1] 432.0 345.6

Here you can see the clusters that are made based on the data of Spotify. Because my entire corpus was too big to incorporate, I narrowed it down to three tracks per album:

When you look at the resulting clusters you can see four distinguishable clusters and a lot of loose tracks or small clusters on the left. The algorithm did quite a good job in forming the bigger clusters. Most of the tracks in the clusters are from the same period or even the same album. This proves there is some similarity between the tracks in certain periods.

If you look at the resulting heatmap you see a cluster that is mostly based on loudness and energy and for example one on acousticness, c06 and c07. There are more patterns visible, but they are vaguer than the ones I mentioned. For example, there is a small cluster around valence, c04 and danceability. The tracks in this cluster are from Davis’s later albums that are more based on popular music, what could explain the danceability

Conclusion

Conclusion

If you add up all the results you could conclude that there certainly is a change in Davis’ style. This change is represented in the data less firm then I expected, but as we saw earlier there is a change in centralization of the tracks overtime. In the first period the tracks are acoustic, while this is completely different at the end of Davis’ career where his music becomes more and more based on popular influences. This is also visible in the audio features; the music becomes more danceable, more often based on a verse-chorus structure (as we saw in the self-similarity matrices) and most important less acoustical.

Although the graphs based on the Spotify features show a change in general, in my opinion the data isn’t very reliable for specific tracks. This goes wrong a lot of times when looking at the acousticness of a track and I doubt if the acousticness feature isn’t just based on the energy and the loudness of a song to a great extent. If Spotify would give us some more information on how this feature is measured it would help a lot to decide whether it is a reliable source or not. For now, I would say that this isn’t the case when you look at specific tracks. As I said before it gives you a good picture of a corpus in general but the dispersion in the plot from tracks on the same album makes me doubt the feature (certainly because I hear something else).

Coming back to the music of Miles Davis in a way it is true he changed music five or six times, but the question is: didn’t he just follow the trends in jazz music? He had an eye for what became the new style and tried to implement it in his own music, something he did quite well. As said in the introduction this was an exaggerated statement, but after looking at the audio features of the music you could clearly speak of a change of style.